Appendix C:Exemplary guideline for the management of the interaction between thiazides (and related diuretics) and selective serotonin reuptake inhibitors (SSRIs) or serotonin-norepinephrine reuptake inhibitors(SNRIs)
Supplementary material to:Checklist for standardized reporting of drug-drug interaction management guidelines, European Journal of Clinical Pharmacology
Authors: Annemieke Floor-Schreudering1,2,Kris L.L. Movig3, Marcel L. Bouvy1,4, Peter A. G. M. De Smet2,5
1SIR Institute for Pharmacy Practice and Policy, Leiden, the Netherlands; 2Departments of Clinical Pharmacy and IQ Healthcare, University Medical Centre St Radboud, Nijmegen, the Netherlands; 3Department of Clinical Pharmacy, Medisch Spectrum Twente, Enschede, the Netherlands; 4Division PharmacoepidemiologyClinical Pharmacology, Utrecht Institute for Pharmaceutical Sciences (UIPS), Utrecht University, Utrecht, the Netherlands; 5Scientific Institute of Dutch Pharmacists (WINAp), the Hague, the Netherlands
Correspondence: Mrs. Annemieke Floor-Schreudering, SIR Institute for Pharmacy Practice and Policy, Theda Mansholtstraat 5B, 2331 JE Leiden, The Netherlands. Tel: +31 71 5766157, Fax: +31 71 5722431, E-mail:
Domain Background (Reason for managing the DDI)1.A. Interaction drugs and drug groups involved in the DDI (only exhaustive for the Dutch drug market)
-Thiazide diuretics: bendroflumethiazide, chlorothiazide, cyclothiazide, hydroflumethiazide, hydrochlorothiazide, and polythiazide.
-Diuretics related to thiazide diuretics:chlortalidone, clopamide, indapamide, mefruside, metolazone, and xipamide.
-Selective serotonin reuptake inhibitors (SSRIs): citalopram, escitalopram, fluoxetine, fluvoxamine, paroxetine, and sertraline.
-Serotonin-norepinephrine reuptake inhibitors (SNRIs): desvenlafaxine, duloxetine, venlafaxine. There is evidence to suggest that the noradrenaline reuptake inhibiting (NRI) potency of these agents is too large to enable substantial NRI effects at therapeutic doses, without engendering excessive serotonin reuptake inhibiting (SRI) effects.
-To err on the safe side of caution, clomipramine should also be included, because this tricyclic antidepressanthas similar SRI potency as SSRIs[1].
2. Clinical impact of DDI at population level[a]
Summary
Both thiazides/related diuretics and SSRIs/SNRIs entail a risk of hyponatraemia (HN), when given alone. The resulting HN is most often mild (sodium 135 mmol/l but ≥130 mmol/l), but both drug groups have been associated with severe HN in case reports [Grade C], which can potentially be fatal [seriousness code F].
Severe HN has been observed in 0.9% of 951 thiazide users in primary care [Grade B] and
in 1 of 75 (1.3%) elderly depressed patients started on paroxetine [Grade C]. It isgenerally assumed that the risk of severe HN can increase further by combining the thiazide with an SSRI/ SNRI but the extent to which this occurs is unknown.A small-sized case control study has provided evidence to suggest that such combinations may have a synergistic effect rather than an additional one [Grade C].
Seriousness of potential adverse outcomes of DDI
The seriousness of the potential adverse outcomes of the DDI between thiazides/related diuretics and SSRI/SNRIs depends on the severity of the hyponatraemia (HN) that can develop. In general, HNis classified as [4,5]:
-mild (serum sodium < or 135 mmol/l but > or ≥ 130 mmol/l);
-moderate (serum sodium < or 130 mmol/l but > or ≥ 125 mmol/l);
-severe (serum sodium < or 125 mmol/l)
Mild to moderate cases of HNassociated with a thiazide/related diuretic (HN-T)often remain asymptomatic, but vaguesymptoms such as fatigue or nausea are possible[6]. Some patients experience gait and attentiondeficits which increase the risk of falls and fractures[7]. Severe HN-T can be associated withnausea and vomiting, abdominal pain, muscularweakness, headache, confusion, dizziness, and lethargy[6,7]. Severe HN-T that is rapidly evolving can lead to cerebral edema, seizures, coma, brain damage, respiratory arrest, brain stemherniation, and death[8-10]. However, severe HN-T apparently has a lower risk of mortality than severe HN by other causes (such as heart failure or liver disease) [11].
Incidence of adverse outcomes in patients using DDI combination
No concrete information has been found in the literature about the incidence of HN in patients given combinations of a thiazide/related diuretic and an SSRI/SNRI.Table 1 summarizes more generalstudies of the incidence of HN(in general or severe) in different populations as well as the risk factors observed in these studies. It should be noted thatthe latter information is determinedby the selection of risk factors in the underlying studies.
Thiazides and related diuretics are a commoncause of severe hyponatraemia[11,12] and are thereby among the more common causes of potentially preventable drug-related hospital admissions[13,14].
In the Dutch Rotterdam Study of elderly outpatients, HN was seen in 3.9% of 13,325 patients ≥ 45 years. The HN was severe in 48 cases (0.4%) and exposure to thiazide diureticssignificantly increased the risk ofsevere HN (HRadj = 8.3) [5,15].In a study in Britishprimarycare, HN was seen in 13.7% of 951 users of thiazides or related diuretics; 9(0.9 %)of the cases were severe[4].
In a prospective study, initiation of paroxetine produced HN in 9 of 75 (12%)elderly patients with depression; the HN was moderate in 6 patients (8%) and severe in 1 (1.3%) patient [16].
Concrete information about the incidence of individual adverse outcomes/symptoms associated with severe HN (such as seizures) has not been retrieved from the literature.
3. Patient susceptibility factors[b]
Summary
Specific studies of the risk factors which increase the susceptibility for severe HN in patients using a combination of a thiazide/related diuretic and an SSRI/SNRI have not been found in the literature. However, considerable information exists on the susceptibility factors which increase the risk of HN in generalorthe risk of HN associated with a thiazide/related diuretic or with an SSRI/SNRI.On the basis of this information, Table 2 summarizes the principal risk factors to be taken into account when evaluating the potential risk of HN when combining a thiazide/related diuretic with an SSRI/SNRI[Grade C].
HN in general
Table 3 lists various causes of HN in general. Table 4 reviews which drugs and drug groups have been associated with HN. These general causes of HN must be taken into consideration in addition to the specific risk factors discussed below.
The risk that sodium and water homeostasis is affected by intercurrent infections, vomiting, diarrhea, fever, great physical strain, or a heat wave should also be considered [13,17,18].
Pre-existing asymptomatic HN is associated with a high risk for the development of severesymptomatic HN[19].
HNassociated with a thiazide/related diuretic (HN-T)
Epidemiological studies and case reports imply that a rechallenge after HN-T, older age, lower body mass, lower serum potassium, and concurrent use of certain other drugs are among the major risk factors for HN-T.The risk can also be increased by a high dose of thiazide, high fluid intake(polydipsia), low sodium intake, and perhaps by a prior serum sodium concentration in the low–normal range.Female gender and impaired renal function have also been implicated but these have not emerged as independent risk factorsin epidemiological studies (cf. Table 1)[6,7,10,20-26].
HN associated with an SSRI/ SNRI (HN-S)
Epidemiological studies and case reports imply that the risk of HN-Sis increased by a rechallenge after HN-S, older age, lower body mass, high serum potassium (>5.0 mmol/l), low sodium intake, and concurrent use of diuretics (cf. Table 1) [27-29].
In a prospective study, initiation of paroxetine produced moderate to severe HN in 7 (9%)of 75 elderly patients with depression, all of whom had a pre-existing sodium level of 135-139 mmol/L[16].
HNassociated with a thiazide/related diuretic plus an SSRI/SNRI
It is generally assumed that combinations of a thiazide/related diuretic with an SSRI/SNRI entail a higher risk of HN than each drug alone [17,22,24,30,31].In a small case control analysis of antidepressant users ≥ 65 years with sodium 130 mmol/l versus users with sodium >135 mmol/l, SSRIs alone and diuretics alone yielded an ORadjof 6.2 and 1.0, respectively, while their combination gave an ORadj of 148[32].
4. Quality of evidence for harm by the DDI[c]
Level of evidence for harm[Grade C]
Published clinical evidence that the combined use of a thiazide/related diuretic and an SSRI/SNRI increasesthe risk of severe symptomatic HN consists of case reports[24,33] and a small-sized case control study[32].
Pharmacological plausibility of the DDI[Plausibility code 1]
There is evidence from epidemiological studies and case reports that thiazides/related diuretics and SSRIs/SNRIs can both produce severe HN, when given alone. Positive rechallenges have been reported for the thiazide diuretics [10,20,34,35] and more rarely for the SSRIs/SNRIs [22,27,28,36-38].
The likelihood of a pharmacodynamic DDI between these drug groups is supported by case reports [24,33] and a small case-control study[32].Another argument is that these drug groups probably induce severe HN through different mechanisms:
-Thiazides/related diuretics
The mechanisms underlying HN-Tare not yet fully understood. The fact that only some individuals develop a severe, dose-independent, HNsuggests an idiosyncratic effect, which perhaps comes from a genetic predisposition[7]. Among the mechanisms suggested so far are[35,39,40]:
-excess renal loss of sodium and potassium (resulting from diuretic-induced electrolyte losses and ADH-induced water retention);
-diuretic-induced volume depletion (which stimulates ADH secretion);
-coexisting hypokalemia leading to an exchange in which potassium leaves the cells to replenish extracellular stores, whereas sodium moves into the cells;
-direct inhibition of urinary dilution by diminishing sodium chloride reabsorption in the
renal tubules;
-stimulation of thirst;
-excessive ADH secretion.
-SSRIs/SNRIs
HN associated with the use of SSRIs/SNRIs is generally attributed to a central increase of ADH secretion[33,39].
Domain Management (How to manage)5. Potentialmanagement strategies for preventing adverse DDI outcome[d]
Strategy I. Do not start drug [grade C]
If one of the two interacting drugs has not yet been started, one may substitute the drug to be started with a non-interacting alternative drug, ifadditional risk factorsare present(Table 2):
-a thiazide or related diuretic may be replaced with another type of antihypertensive;
-a SSRI or SNRI may be replaced with an another type of antidepressant such as mirtazapine or a tricyclic antidepressant other than clomipramine (cf. Table 4). Although such alternatives have been associated less frequently with HN, it should be noted that they may not always be risk free in patients susceptible to SSRI/SNRI-associated HN [41].
Substitution is especially the most appropriate course of action, if the drug to be started has already produced HN in the patient in the past. It has been reported that rechallenge of susceptible patients with a single dose of a thiazide can be enough for an acute drop of serum sodium concentration[20].
Strategy II. Monitoring of the patient [Grade C]
There is no consensus in the literature how useful determination of the serum sodium concentration is in asymptomatic patients[30].A common recommendation is to determine serum sodium at 5-14 days after the start of the combination only in users with one or more additional risk factors (Table 2) [6,7,14,17,31].A repeat measurementis appropriate, when the risk is further increased by an additional risk factor (e.g., intercurrent illness) or by an increase in the dose of the thiazide/related diuretic or SSRI/SNRI [14].
When to monitor
Before treatment, it should be established if the patient has additional risk factors (Table 2).
If additional risk factors are present, the patient should be monitored for symptoms of HN and/or for a subnormal serum sodium concentration. This is particularly relevant in the first 2 weeks of combined treatment (when the risk of developing HN is the highest).
Monitoring for symptoms of HN and/or for a subnormal serum sodium concentration may also be warranted at a later stage (e.g., after an increase in dose or when the existing risk is aggravated by a temporary risk factor, such as an intercurrent infection or heat wave). In a primary care study of serum sodium concentration in thiazide users, 20% of the HN cases were not detected by thefirst electrolyte check but on subsequent sampling[4].
Cost-effectiveness of monitoring
Dutch societal cost of a potentially preventable drug-related hospital admission was estimated at € 4573 in 2006.[42] Applying an inflation correction factor of 1.154[43],the corresponding figure in 2013 is € 5277. The cost of a sodium laboratory test in primary care is € 1.96 in 2013 [44].This means that sodium testing of patients on a thiazide/related diuretic + SSRI/SNRI is already cost-effective if one HN-related hospitalization is prevented by less than 2692 sodium tests.
Response to HN
The response to HN depends on the serum sodium concentration, the presence ofsymptoms and the reasons for prescribing an SSRI/SNRI and a thiazide/related diuretic [4,6,7,25,45]:
-Patients with mild to moderate HN are likely to be asymptomatic or minimally symptomatic (nausea, vomiting, cognitive impairment, confusion), If possible, the thiazide or related diuretic should be replaced with an alternative antihypertensive drug and additional risk factors should also be corrected (e.g., by restricting water intake to less than 1 L/day). These measures are likely to be sufficient for the normalization of the sodium concentration within 1–5 days.
If blood pressure cannot be controlled without the aid of a thiazide, replacement of the SSRI/SNRI with another antidepressant seems the most viable option. Alternatively, one could provisionally continue the drug combination under careful further monitoring of the sodium concentration.
-Patients with severe HN are likely to be symptomatic and often have to be submitted to specialized care in hospital, where a delicate balance has to be found between the risk of cerebral edema and the risk of an osmotic demyelination syndrome due to overly rapid correction of the HN[6,12].See the articles by Liamis et al. 2006, Gross 2008, Hoorn et al. 2008, and Vaidya et al. 2010 for further details [45-48].
In the primary care study by Clayton et al. 2005 [4], all except one of thepatients who continued thiazide treatment despite moderate or severe HN remainedhyponatremic, suggesting that spontaneous normalizationis unlikely if the thiazide is continued.
Strategy III. Consultation and instruction of the patient[Grade C]
Patients at increased risk of HN (or their carers) should be instructed orally and in written form to consult with their prescriber, if they start to experience HN-like symptoms (such as fatigue, nausea,and vomiting, abdominal pain, muscularweakness,gait and attentiondeficits, headache, confusion, dizziness, and lethargy).
Patients at increased risk should be warned about the risk of excessive fluid intake and they should be informed about which situations increase the risk of sodium and water loss (e.g. intercurrent infections, vomiting, diarrhea, fever, great physical strain, or a heat wave (see above). Vulnerable elderly patients may require additional guidance in such situations.
The following caveats should be taken into consideration, when the patient or carer is informed about the risk of HN:
-the first symptoms of HN are not specific and may develop slowly, both of which can hamper timely recognition [17].
-the information should be presented very carefully (preferably in oralplus written form) to prevent that the patient is so frightened that he decides on his own not to take the prescribed riskmedication[13].
6. Summary[e]
Background (Reason for managingthe DDI)
Both thiazides/related diuretics and SSRIs/SNRIs entail a risk of hyponatraemia (HN), when given alone. The resulting HN is most often mild (sodium 135 mmol/l but ≥130 mmol/l ), but both drug groups have been associated with severe HN in case reports [Grade C], which can potentially be fatal [seriousness code F].
Severe HN has been observed in 0.9% of 951 thiazide users in primary care [Grade B] and
in 1 of 75 (1.3%) elderly depressed patients started on paroxetine [Grade C]. It isgenerally assumed that the risk of severe HN can increase further by combining the thiazide with an SSRI/ SNRI but the extent to which this occurs is not known.A small-sized case control study has provided evidence to suggest that such combinations may have a synergistic effect rather than an additional one [Grade C].
A pragmatic selection of major patient susceptibility factors is presented in Table 2 [Grade C].
Management strategy (How to manage)
The different options for management of the DDI between a thiazide/related diuretic (T) and a Selective Serotonin Reuptake Inhibitor/Serotonin-Norepinephrine Reuptake Inhibitor (S)in a patient who is startedon T and/or S are summarized in the algorithm below.
The first step is to establish the presence or absence of additional risk factors before the DDI is started (Table 2):
-When additional risk factors are absent, the only action to be considered is consultation and instruction of the patient [Grade 2-C].
-When one or more additional risk factors are present (but the patient has not experienced HN associated with a thiazide or SSRI/SNRIs in the past), one may consider the option to change the newly prescribed drug [Grade 2-C]. If this is not the preferred option the sodium concentration of the patientshould be measured before and within the first 2 weeks after the start of the DDI [Grade 1-C] and the patient should be consulted/instructed [Grade 2-C].
-If the patient has already experienced HN associated with a thiazide/related diuretic or an SSRI/SNRI in the past, the best action is to avoid drugs from the incriminated drug class [Grade 1-C].After substitution of the incriminated drug, the patientshould be measured before and within 2 weeks after the start of the DDI [Grade 1-C] and the patient should be consulted/instructed [Grade 2-C].
As there are no well-designed studies to support the effectiveness of the proposed interventions, the evidence underlying each recommendation for intervention has each time been rated as weak [Grade C]. In view of the potential seriousness of the adverse outcome, some recommendations have nevertheless been graded as strong:
-do not start a thiazide/related diuretic or SSRI/SNRI, if the patient has alreadyexperienced HN associated with that particular drug class in the past [Grade 1-C]
-monitor the sodium concentration of the patient before and just after the start of the DDI if the patient has additional risk factors for HN [Grade 1-C].
Algorithm for management of drug-drug interaction between thiazide/related diuretic (T) and Selective Serotonin Reuptake Inhibitor/Serotonin-Norepinephrine Reuptake Inhibitor (S) in patient started on T and/or S1
Tables
Table 1. Risk size and potential risk factors of hyponatraemia in epidemiological studies (*)Study Population / Hyponatraemia (HN) / Potential Risk Factors
Reference(s) / Patients and Setting / Overall / Severe / Significant Associations
[OR, RR, HR] / Not significant
Outpatients in general
[5,15] / 13,325 Dutch outpatients ≥ 45 years
(Rotterdam Study) / 3.9% / 0.4% / Thiazides [HRadj = 4.9] overall)
[HRadj = 8.3] (severe) / Gender
[49] / 5179 Dutch outpatients > 55 years
(Rotterdam Study) / Diabetes mellitus [OR = 2.0]
Thiazides [OR = 1.7]
Potassium-sparing diuretics [OR = 3.6] Antiepileptics [OR = 3.0]
Benzodiazepines [OR = 1.4] / Heart failure
Loop diuretics
Nursing home residents in general
[50] / 119 nursing home residents vs 60 subjects attending geriatric outpatient clinic. / 18% / High intake of hypotonic fluid
Low sodium content of tube feeding
Spinal cord diseases
Inpatients with hyponatraemia (HN) in general
[51] / 223 thiazide users in Hong Kong hospitalized with HN (sodium < 130 mmol/l) vs thiazide users without HN / Higher age (76 vs 66 years)
Lower body weight (52 vs 63 kg)
Lower serum potassium (3.4 vs 4.0 mmol/l) / Gender
Renal function
Duration of thiazide use
Loop diuretics
ACE inhibitors
NSAIDs
[52] / 50 cases with hospital-acquired HN vs 69 matched controls without HN / 3.3% / Insulins [ORadj = 10.5]
Antibiotics [ORadj = 4.5]
Opioids [ORadj = 2.9]
[53] / 93 psychiatric inpatients with HN (<130 mmol/l) vs 263 864 psychiatric inpatients / 0.04% / Female gender
Age ≥ 65years
SSRI + diuretic or ACE inhibitor
SSRI + diuretic + ACE inhibitor
Venlafaxine + diuretic or ACE inhibitor
Venlafaxine + diuretic + ACE inhibitor
Oxcarbazepine (alone) / SSRI alone
Venlafaxine alone
Users of thiazides
[54] / 2255 hypertensive outpatients ≥ 60 years treated with 12.5-25 mg chlorthalidone (CT) per day
Vs 2189 outpatients on placebo / 4.1% (CT)
1.3% (P)
(sodium 130 mmol/l) ($)
[4] / 951 outpatients in UK primary care / 13.7% / 0.9% / Age > 70 years [OR = 3.9] / Gender
Dose level of thiazide
[5,15] / 3,456 Dutch outpatients ≥ 45 years
(Rotterdam Study) / 4.9% / Lower BMI
Younger age / Gender
Renal function
[55] / 220 outpatients in US / 30% (sodium 130 mmol/l) ($)
Users of SSRIs/SNRIs
[56] / 845 SSRI users ≥ 65 years in New Zealand / 1.6%
[57] / Korean users of antidepressants:
93 SSRI
71 venlafaxine (V)
76 mirtazapine (M) / 8.6% (SSRI)
4.2% (V)
0% (M)
[56] / 14 SSRI users with HN vs 56 SSRI users without HN / Body mass [OR = 0.9]
Age [OR = 1.1] / Gender
Renal function
[32] / 29 antidepressant users with sodium 130 mmol/l vs 78 antidepressant users with sodium >135 mmol/l) / SSRIs [OR= 3.3]
Potassium >5.0 mmol/l [OR = 16]
SSRIs + diuretics + age ≥65 years [OR = 13.5]
[22] / 74 inpatients > 65 years using SSRI or venlafaxine vs 125 control patients not using these drugs / 39% / SSRIs or venlafaxine [ORadj = 3.5]
[16] / 75 elderly patients (aged 63-90 years) started on paroxetine / 12% / 1.3% / Low body mass index
Low baseline sodium level (<138 mEq/L)
[58] / HN in 60,746 depressed patients ≥ 65 years / Absolute risk over 1 year
0.44% (SSRIs)
0.30% tricyclic antidepressant
0.29% no antidepressant / SSRIs [HRadj = 1.5 when compared to no use of antidepressant] / Tricyclic antidepressants
(*)hyponatraemia= HN; see for definitions [4] and [5]: